•Achieved a 45% decrease in model size and 2/3 decrease in model inference time by applying PyTorch and Onnx quantization to user score prediction models within a language learning app for English exams with 3MM+ global users.
•Managed CI/CD of the AI inference server, A/B test dashboard, and OneFlow repositories and implemented integration test automation with lint and test coverage using Github Action to decrease overall time for model deployment.
•Developed an in-company dashboard for Riiid that displays company product info, market statistics (KPIs), and HR information using Typescript, React.js, Next.js, Vercel, Airflow, and Notion/Asana/Google Sheets API.
•Built an in-company platform that enabled 150 users at Riiid to express opinions on specific parts of online/offline meetings with a web app and helped optimize content for presentations using Typescript, React.js, Next.js, Vercel, and Firebase.